Lutheran Drafts: Which Draft Position Do You Want, And How Do You Get It?

Most Fantasy drafts embody a sort of Calvinist view of the world: your draft position is a matter of predestination, where you wind up is arbitrarily determined, and there’s nothing you can do to alter the outcome. But drafts in the National Fantasy Baseball League are more Lutheran: there are things you can do to affect your position. To determine draft order in snake drafts, the NFBC uses what it calls the Kentucky Derby System, because it resembles the way post positions get chosen for the Derby. NFBC owners can indicate their draft position preferences beforehand by ranking them. If the owner doesn’t bother, the default ranking for that team is what you’d expect: 1 through Whatever. The NFBC computer then randomly picks the order in which each owner’s preferences are consulted. The first owner gets her first choice, the second owner gets his first choice unless it’s already gone, in which case the computer moves on to the next owner and doesn’t come back to Owner Number Two until everyone else’s first preferences have been consulted. And so it proceeds with second preferences, third preferences, and so on. Thus, it’s theoretically possible that the last owner in the KDS sequence gets the first draft choice.

The question is, does she want it? And that’s what we decided to find out: are there any differences at all, this year, among draft positions? If so, how big are the differences and which positions are best? And how can you go about getting those positions?

Here’s what we did: We started with Ariel Cohen’s ATC 2018 stat projections, which you will find in the Projections section on the Fangraphs site. (We could of course have used Steamer or anyone else’s projections, but we like Ariel’s.) To determine the dollar values of each player’s package of stats, we used our homemade Standings Gain Points. (We won’t digress to explain SGPs; the explanation to be found on Tanner Bell’s excellent web site will tell you what you need to know if you don’t know it already.) Forgive the hubris, but we prefer our SGPs to whatever Fangraphs uses to calculate auction values on its Auction Calculator. We also use them in lieu of other people’s: they’re similar, but the denominators are smaller because we eliminate extreme outcomes at either end. We also used a 67/33 hitter/pitcher split, which is a bit more for hitters and less for pitchers than seems currently fashionable. It looks to us, from running the Auction Calculator with different sets of projections and different hitter/pitcher splits and eyeballing the results, that the results we report below aren’t aberrational.

We computed dollar values for all the players Ariel projects, and then took the top 210 hitters and 145 pitchers—the players needed to assemble 23-man rosters in 15-team leagues. We adjusted for position scarcity, which is a necessary tweak to account for the melancholy fact that catchers generally don’t play every day or hit very well when they do play. The only roster constraint we introduced at this point was to require that each team have two “closers,” which we defined as pitchers whom Ariel projected for 10 or more saves. The other seven pitchers could be starters or non-closer relievers. Then we conducted a mock snake draft in which the players were taken in order of their dollar value until each roster was complete, and calculated the total dollar value of each roster. Par, we figured, would be $260, the amount of each team’s budget in a standard Rotisserie auction. The results (DP is Draft Position):

DP 1: $267.97
DP 2: $265.13
DP 3: $263.33
DP 4: $259.99
DP 6: $259.82
DP 9: $259.76
DP 7: $259.64
DP 8: $259.60
DP 5: $259.51
DP 11: $259.47
DP 10: $258.27
DP 12: $257.82
DP 13: $257.17
DP 15: $256.37
DP 14: $256.12

Big difference between the top 3 and everyone else, right? So we wondered if the difference was structural, i.e. endemic to the draft process itself, or circumstantial, i.e. attributable to value differences among specific players. Here are the players taken in the first two rounds and the total dollar values of those players:

DP 1:$66.75 (Trout, Martinez)
DP 2:$64.22 (Turner, Strasburg)
DP 3:$63.03 (Altuve, Marte)
DP 6:$60.29 (Goldschmidt, Correa)
DP 9:$60.14 (Sale, Machado)
DP 7:$60.07 (Blackmon, Carrasco)
DP 8:$60.01 (Betts, Bellinger)
DP 11:$59.98 (Scherzer, Judge)
DP 5:$59.79 (Arenado, Dozier)
DP 4:$59.51 (Kershaw, Rizzo)
DP 10:$58.77 (Kluber, Harper)
DP 12:$58.38 (Gordon, Sanchez)
DP 13:$57.86 (Stanton, Lindor)
DP 14:$57.10 (Bryant, Freeman)
DP 15:$57.09 (Ramirez, Votto)

And here are the dollar values for the remaining rounds:

DP 1: $201.22
DP 2: $200.91
DP 4: $200.47
DP 3: $200.30
DP 5: $199.72
DP 9: $199.62
DP 8: $199.59
DP 7: $199.56
DP 6: $199.53
DP 10: $199.50
DP 11: $199.50
DP 12: $199.54
DP 13: $199.30
DP 15: $199.28
DP 14: $199.03

So once you take those first two rounds out, the values are much closer to identical, though there’s still a small first-to-last decline.

To get a clearer picture of this, we needed to make one more adjustment. Up to this point, we hadn’t adjusted for positions. It would have been possible, say, for a team to have nine outfielders. (Why did we do it this way? Partly because the only comparable study we’ve heard of did it this way, and partly because it was a whole lot easier than doing what we’re about to describe.) So we conducted an actual draft, using the positional constraints of the classic 23-man Rotisserie roster, which is also what the NFBC uses for its Main Event. Each team took the highest-priced player available for whom it had room. Here’s the outcome:

DP 1: $268.67
DP 2: $266.06
DP 3: $263.22
DP 5: $261.22
DP 4: $260.24
DP 7: $260.01
DP 6: $259.77
DP 8: $259.57
DP 9: $259.49
DP 11: $259.19
DP 10: $258.59
DP 12: $257.05
DP 15: $255.93
DP 13: $255.67
DP 14: $255.09

Same differences, somewhat magnified. And when you remove the first two picks from the calculation, there’s the same partial smoothing but the same first-to-last decline we saw before:

DP 1: $201.92
DP 2: $201.84
DP 5: $201.62
DP 4: $200.73
DP 3: $200.19
DP 7: $199.94
DP 10: $199.82
DP 8: $199.56
DP 6: $199.48
DP 9: $199.35
DP 11: $199.21
DP 15: $198.84
DP 12: $198.67
DP 14: $197.99
DP 13: $197.81

A few notes before we discuss strategy: First, it would be interesting to know whether the same differences appear in other seasons. For example: using plausible projections and SGPs for 2017, would these have been the results? Perhaps not exactly. Our values had Trout/Turner/Altuve a good bit higher than anyone else, whereas last year we seem to recall that only Trout and Kershaw separated themselves in the bidding, with a cluster of four or five guys in the next tier. In other words, maybe the difference is circumstantial rather than structural, and the same analysis next season will produce a different set of preferences. But if you run the Fangraphs Auction Calculator with different projections and different hitter/pitcher allocations, you’ll get some different players on top and different price distributions, but (it appears) similar outcomes. And even after removing the picks in the first two rounds, there’s a small but real round-by-round difference between earlier choices and later choices. Looking at quads of choices later in the draft, we see that you continue to do better with an early pick in the odd rounds and a later pick in the even rounds than vice versa. Here’s the total value of picks 3 through 6 for each team:

DP 1: $78.28
DP 2: $78.01
DP 4: $77.40
DP 3: $77.39
DP 11: $77.19
DP 12: $77.13
DP 15: $77.11
DP 14: $77.07
DP 10: $77.03
DP 13: $77.03
DP 9: $76.92
DP 8: $76.84
DP 7: $76.80
DP 5: $76.71
DP 5: $76.66

And here’s the total value of picks 7 through 10:

DP 1: $50.86
DP 3: $50.83
DP 2: $50.79
DP 4: $50.79
DP 5: $50.70
DP 6: $50.69
DP 7: $50.67
DP 8: $50.64
DP 9: $50.52
DP 10: $50.44
DP 13: $50.25
DP 11: $50.16
DP 12: $50.16
DP 14: $50.07
DP 15: $50.05

And of course beyond that, in a draft where the roster must be filled in the 23rd round, there’s a built-in advantage to going first in that round, since there’s no 24th round to balance things out. The first choice in the 23rd round was worth about 30 cents more than the last choice. So perhaps the difference is structural, though the exact architecture may differ somewhat from year to year.

So now you know what you already thought you knew: that you’d rather pick first than last and earlier than later. The question is, what do you do with this knowledge? The most obvious takeaway is that the “wheel” positions 13-14-15, where you get to make two picks at a time in close proximity) are a snare and a delusion. Whatever benefit they yield in terms of planning is probably outweighed by the detriment of drafting that late in the odd rounds. Beyond that, there appear to be three rough clusters of positions: 1-3, 4-11 or -12, and 12- or 13-15. In the absence of any reason to think otherwise, you have to assume that everyone else is going to ask for 1 through 15 in that order. If you do the same, you’re back in the Calvinist cosmos. Maybe you’re better off, then, making 2 or (probably even better) 3 your top choice and 5,6, or even 11 your second. And if you’re very risk-averse, you could put 5, 6, or 11 on top.